fix(api): Do not call .plot() in display factory (#1299)

closes #1298 

This PR makes sure that the display factory does not call `.plot()`
directly.
It allows:

- to pass plotting parameters to be passed only to `display.plot()`
- to have a consistent api as for pandas object like dataframe:

```python
report.metrics.report_metrics().plot(kind="barh")
report.metrics.roc().plot()
```

examples/getting_started/plot_skore_getting_started.py

@@ -77,7 +77,7 @@
 import matplotlib.pyplot as plt
 
 roc_plot = est_report.metrics.roc()
-roc_plot
+roc_plot.plot()
 plt.tight_layout()
 
 # %%
@@ -117,7 +117,7 @@
 
 # %%
 roc_plot_cv = cv_report.metrics.roc()
-roc_plot_cv
+roc_plot_cv.plot()
 plt.tight_layout()
 
 # %%

examples/model_evaluation/plot_estimator_report.py

@@ -335,6 +335,7 @@ def operational_decision_cost(y_true, y_pred, amount):
 #
 # Let's start by plotting the ROC curve for our binary classification task.
 display = report.metrics.roc(pos_label=pos_label)
+display.plot()
 plt.tight_layout()
 
 # %%
@@ -362,7 +363,8 @@ def operational_decision_cost(y_true, y_pred, amount):
 # performance gain we can get.
 start = time.time()
 # we already trigger the computation of the predictions in a previous call
-report.metrics.roc(pos_label=pos_label)
+display = report.metrics.roc(pos_label=pos_label)
+display.plot()
 plt.tight_layout()
 end = time.time()
 
@@ -376,7 +378,8 @@ def operational_decision_cost(y_true, y_pred, amount):
 
 # %%
 start = time.time()
-report.metrics.roc(pos_label=pos_label)
+display = report.metrics.roc(pos_label=pos_label)
+display.plot()
 plt.tight_layout()
 end = time.time()
 

examples/use_cases/plot_employee_salaries.py

@@ -313,10 +313,12 @@ def periodic_spline_transformer(period, n_splines=None, degree=3):
     if estimator_report is None:
         ax.axis("off")
         continue
-    estimator_report.metrics.prediction_error(kind="actual_vs_predicted", ax=ax)
+    estimator_report.metrics.prediction_error().plot(kind="actual_vs_predicted", ax=ax)
     ax.set_title(f"Split #{split_idx + 1}")
     ax.legend(loc="lower right")
 plt.tight_layout()
 # sphinx_gallery_start_ignore
 temp_dir.cleanup()
 # sphinx_gallery_end_ignore
+
+# %%

skore/src/skore/sklearn/_cross_validation/metrics_accessor.py

@@ -812,7 +812,6 @@ def _get_display(
         response_method,
         display_class,
         display_kwargs,
-        display_plot_kwargs,
     ):
         """Get the display from the cache or compute it.
 
@@ -833,9 +832,6 @@ def _get_display(
         display_kwargs : dict
             The display kwargs used by `display_class._from_predictions`.
 
-        display_plot_kwargs : dict
-            The display kwargs used by `display.plot`.
-
         Returns
         -------
         display : display_class
@@ -852,7 +848,6 @@ def _get_display(
 
         if cache_key in self._parent._cache:
             display = self._parent._cache[cache_key]
-            display.plot(**display_plot_kwargs)
         else:
             y_true, y_pred = [], []
             for report in self._parent.estimator_reports_:
@@ -882,7 +877,6 @@ def _get_display(
                 ml_task=self._parent._ml_task,
                 data_source=data_source,
                 **display_kwargs,
-                **display_plot_kwargs,
             )
             self._parent._cache[cache_key] = display
 
@@ -893,7 +887,7 @@ def _get_display(
             supported_ml_tasks=["binary-classification", "multiclass-classification"]
         )
     )
-    def roc(self, *, data_source="test", pos_label=None, ax=None):
+    def roc(self, *, data_source="test", pos_label=None):
         """Plot the ROC curve.
 
         Parameters
@@ -907,9 +901,6 @@ def roc(self, *, data_source="test", pos_label=None, ax=None):
         pos_label : int, float, bool or str, default=None
             The positive class.
 
-        ax : matplotlib.axes.Axes, default=None
-            The axes to plot on.
-
         Returns
         -------
         RocCurveDisplay
@@ -928,21 +919,19 @@ def roc(self, *, data_source="test", pos_label=None, ax=None):
         """
         response_method = ("predict_proba", "decision_function")
         display_kwargs = {"pos_label": pos_label}
-        display_plot_kwargs = {"ax": ax, "plot_chance_level": True, "despine": True}
         return self._get_display(
             data_source=data_source,
             response_method=response_method,
             display_class=RocCurveDisplay,
             display_kwargs=display_kwargs,
-            display_plot_kwargs=display_plot_kwargs,
         )
 
     @available_if(
         _check_supported_ml_task(
             supported_ml_tasks=["binary-classification", "multiclass-classification"]
         )
     )
-    def precision_recall(self, *, data_source="test", pos_label=None, ax=None):
+    def precision_recall(self, *, data_source="test", pos_label=None):
         """Plot the precision-recall curve.
 
         Parameters
@@ -956,9 +945,6 @@ def precision_recall(self, *, data_source="test", pos_label=None, ax=None):
         pos_label : int, float, bool or str, default=None
             The positive class.
 
-        ax : matplotlib.axes.Axes, default=None
-            The axes to plot on.
-
         Returns
         -------
         PrecisionRecallCurveDisplay
@@ -977,22 +963,18 @@ def precision_recall(self, *, data_source="test", pos_label=None, ax=None):
         """
         response_method = ("predict_proba", "decision_function")
         display_kwargs = {"pos_label": pos_label}
-        display_plot_kwargs = {"ax": ax, "despine": True}
         return self._get_display(
             data_source=data_source,
             response_method=response_method,
             display_class=PrecisionRecallCurveDisplay,
             display_kwargs=display_kwargs,
-            display_plot_kwargs=display_plot_kwargs,
         )
 
     @available_if(_check_supported_ml_task(supported_ml_tasks=["regression"]))
     def prediction_error(
         self,
         *,
         data_source="test",
-        ax=None,
-        kind="residual_vs_predicted",
         subsample=1_000,
         random_state=None,
     ):
@@ -1008,20 +990,6 @@ def prediction_error(
             - "test" : use the test set provided when creating the report.
             - "train" : use the train set provided when creating the report.
 
-        ax : matplotlib axes, default=None
-            Axes object to plot on. If `None`, a new figure and axes is
-            created.
-
-        kind : {"actual_vs_predicted", "residual_vs_predicted"}, \
-                default="residual_vs_predicted"
-            The type of plot to draw:
-
-            - "actual_vs_predicted" draws the observed values (y-axis) vs.
-              the predicted values (x-axis).
-            - "residual_vs_predicted" draws the residuals, i.e. difference
-              between observed and predicted values, (y-axis) vs. the predicted
-              values (x-axis).
-
         subsample : float, int or None, default=1_000
             Sampling the samples to be shown on the scatter plot. If `float`,
             it should be between 0 and 1 and represents the proportion of the
@@ -1045,17 +1013,13 @@ def prediction_error(
         >>> X, y = load_diabetes(return_X_y=True)
         >>> regressor = Ridge()
         >>> report = CrossValidationReport(regressor, X=X, y=y, cv_splitter=2)
-        >>> display = report.metrics.prediction_error(
-        ...     kind="actual_vs_predicted"
-        ... )
-        >>> display.plot(line_kwargs={"color": "tab:red"})
+        >>> display = report.metrics.prediction_error()
+        >>> display.plot(kind="actual_vs_predicted", line_kwargs={"color": "tab:red"})
         """
         display_kwargs = {"subsample": subsample, "random_state": random_state}
-        display_plot_kwargs = {"ax": ax, "kind": kind}
         return self._get_display(
             data_source=data_source,
             response_method="predict",
             display_class=PredictionErrorDisplay,
             display_kwargs=display_kwargs,
-            display_plot_kwargs=display_plot_kwargs,
         )

skore/src/skore/sklearn/_estimator/metrics_accessor.py

@@ -1396,7 +1396,6 @@ def _get_display(
         response_method,
         display_class,
         display_kwargs,
-        display_plot_kwargs,
     ):
         """Get the display from the cache or compute it.
 
@@ -1424,9 +1423,6 @@ def _get_display(
         display_kwargs : dict
             The display kwargs used by `display_class._from_predictions`.
 
-        display_plot_kwargs : dict
-            The display kwargs used by `display.plot`.
-
         Returns
         -------
         display : display_class
@@ -1442,7 +1438,6 @@ def _get_display(
 
         if cache_key in self._parent._cache:
             display = self._parent._cache[cache_key]
-            display.plot(**display_plot_kwargs)
         else:
             y_pred = _get_cached_response_values(
                 cache=self._parent._cache,
@@ -1463,7 +1458,6 @@ def _get_display(
                 ml_task=self._parent._ml_task,
                 data_source=data_source,
                 **display_kwargs,
-                **display_plot_kwargs,
             )
             self._parent._cache[cache_key] = display
 
@@ -1474,7 +1468,7 @@ def _get_display(
             supported_ml_tasks=["binary-classification", "multiclass-classification"]
         )
     )
-    def roc(self, *, data_source="test", X=None, y=None, pos_label=None, ax=None):
+    def roc(self, *, data_source="test", X=None, y=None, pos_label=None):
         """Plot the ROC curve.
 
         Parameters
@@ -1497,9 +1491,6 @@ def roc(self, *, data_source="test", X=None, y=None, pos_label=None, ax=None):
         pos_label : int, float, bool or str, default=None
             The positive class.
 
-        ax : matplotlib.axes.Axes, default=None
-            The axes to plot on.
-
         Returns
         -------
         RocCurveDisplay
@@ -1527,15 +1518,13 @@ def roc(self, *, data_source="test", X=None, y=None, pos_label=None, ax=None):
         """
         response_method = ("predict_proba", "decision_function")
         display_kwargs = {"pos_label": pos_label}
-        display_plot_kwargs = {"ax": ax, "plot_chance_level": True, "despine": True}
         return self._get_display(
             X=X,
             y=y,
             data_source=data_source,
             response_method=response_method,
             display_class=RocCurveDisplay,
             display_kwargs=display_kwargs,
-            display_plot_kwargs=display_plot_kwargs,
         )
 
     @available_if(
@@ -1550,7 +1539,6 @@ def precision_recall(
         X=None,
         y=None,
         pos_label=None,
-        ax=None,
     ):
         """Plot the precision-recall curve.
 
@@ -1574,9 +1562,6 @@ def precision_recall(
         pos_label : int, float, bool or str, default=None
             The positive class.
 
-        ax : matplotlib.axes.Axes, default=None
-            The axes to plot on.
-
         Returns
         -------
         PrecisionRecallCurveDisplay
@@ -1604,15 +1589,13 @@ def precision_recall(
         """
         response_method = ("predict_proba", "decision_function")
         display_kwargs = {"pos_label": pos_label}
-        display_plot_kwargs = {"ax": ax, "despine": True}
         return self._get_display(
             X=X,
             y=y,
             data_source=data_source,
             response_method=response_method,
             display_class=PrecisionRecallCurveDisplay,
             display_kwargs=display_kwargs,
-            display_plot_kwargs=display_plot_kwargs,
         )
 
     @available_if(_check_supported_ml_task(supported_ml_tasks=["regression"]))
@@ -1622,8 +1605,6 @@ def prediction_error(
         data_source="test",
         X=None,
         y=None,
-        ax=None,
-        kind="residual_vs_predicted",
         subsample=1_000,
         random_state=None,
     ):
@@ -1648,20 +1629,6 @@ def prediction_error(
             New target on which to compute the metric. By default, we use the target
             provided when creating the report.
 
-        ax : matplotlib axes, default=None
-            Axes object to plot on. If `None`, a new figure and axes is
-            created.
-
-        kind : {"actual_vs_predicted", "residual_vs_predicted"}, \
-                default="residual_vs_predicted"
-            The type of plot to draw:
-
-            - "actual_vs_predicted" draws the observed values (y-axis) vs.
-              the predicted values (x-axis).
-            - "residual_vs_predicted" draws the residuals, i.e. difference
-              between observed and predicted values, (y-axis) vs. the predicted
-              values (x-axis).
-
         subsample : float, int or None, default=1_000
             Sampling the samples to be shown on the scatter plot. If `float`,
             it should be between 0 and 1 and represents the proportion of the
@@ -1694,19 +1661,15 @@ def prediction_error(
         ...     X_test=X_test,
         ...     y_test=y_test,
         ... )
-        >>> display = report.metrics.prediction_error(
-        ...     kind="actual_vs_predicted"
-        ... )
+        >>> display = report.metrics.prediction_error()
         >>> display.plot(line_kwargs={"color": "tab:red"})
         """
         display_kwargs = {"subsample": subsample, "random_state": random_state}
-        display_plot_kwargs = {"ax": ax, "kind": kind}
         return self._get_display(
             X=X,
             y=y,
             data_source=data_source,
             response_method="predict",
             display_class=PredictionErrorDisplay,
             display_kwargs=display_kwargs,
-            display_plot_kwargs=display_plot_kwargs,
         )